基于机器学习对串联排队系统等待时间的预测

Prediction of Waiting Time in Tandem Queueing Systems Based on Machine Learning

串联排队系统是排队网络的基本结构,研究串联排队系统对分析排队网络具有重要意义.串联排队系统中站与站之间存在关联性,上游站的输出过程是下游站的输入过程,对于不满足马尔可夫性的排队系统,下游站的到达过程很难用解析的方法分析.对于一般的串联排队系统,本文基于机器学习对串联排队系统的平均等待时间进行预测,通过数值实验比较机器学习中线性回归模型和非线性回归模型的预测效果.实验结果表明,非线性回归模型优于线性回归模型,XGBoost算法对串联排队系统的平均等待时间的预测准确度较高.此外,本文将XGBoost算法与传统的近似分析方法进行比较,发现XGBoost算法的预测效果优于传统的近似分析方法.

Tandem queueing system is the basic structure of queueing network,and it is important to study the tandem queueing system to analyze the queueing network.There are correlations between stations in a tandem queueing system,and the output process of the upstream station is the input process of the downstream station.For queueing systems that do not satisfy Markovianity,the arrival process of the downstream station is difficult to analyze by the analytical method.For the general tandem queueing system,this paper proposes to predict the mean waiting time of the system based on machine learning,and compare the prediction effect of linear regression model and nonlinear regression model in machine learning through numerical experiments.The experimental results show that the nonlinear regression model outperforms the linear regression model,and the XGBoost algorithm has a higher accuracy in predicting the mean waiting time of the tandem queueing system.In addition,this paper compares the XGBoost algorithm with the traditional approximate analysis method and finds that the prediction effect of the XGBoost algorithm is better than that of the traditional approximate analysis method.